home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!dtix!darwin.sura.net!mips!swrinde!cs.utexas.edu!ut-emx!ibmchs!auschs!awdprime.austin.ibm.com!ski.austin.ibm.com!peter
- From: peter@ski.austin.ibm.com (Peter Jeffe 512.838.4019)
- Newsgroups: comp.unix.aix
- Subject: Re: Named spinning wheels
- Message-ID: <1992Aug13.022511.16252@awdprime.austin.ibm.com>
- Date: 13 Aug 92 02:25:11 GMT
- References: <1992Aug3.183040.29775@olaf.wellesley.edu>
- Sender: news@awdprime.austin.ibm.com (USENET News)
- Organization: The Kernel Group, Inc.
- Lines: 30
-
- In article <1992Aug3.183040.29775@olaf.wellesley.edu> kwolters@olaf.wellesley.edu (Keith Wolters) writes:
- >I just installed AIX 3.2.2 on our model 520 which happens to be the
- >nameserver for our campus. named is now doing strange things. It is
- >running but appears to be spinning its wheels. If I do a
- >
- >kill -USR1 pid
- >
- >to turn on debugging, the named.run file grows by 10Kbytes per second.
- >The file caontains the following message over and over.
- >
- >resp: no addrs found for NS's
-
- This is a bug that is present in all versions of BIND that I have seen.
- It occurs when there is an NS record to forward a query to, but there
- is no corresponding A record for the NS; ns_forw() doesn't handle this,
- and goes into a loop. This typically only happens when the A record
- goes stale before the NS record.
-
- But in AIX 3.2 an enhancement to implement caching of temporary failures
- (as per rfc-1123 section 6.1.3.3) has caused this bug to surface much more
- frequently. This is because a server that times out or errors out is
- marked as unavailable for 5 minutes; this causes the A record search in
- nslookup() to fail, which produces the loop.
-
- The fix for this is apar ix23937, which has been packaged for 3.2 in
- ptf U406316, which should be available very soon. For 3.1, emergency
- fixes are currently available.
-
- --
- peter jeffe peter@aixwiz.austin.ibm.com
-